VistaCNC P4-SE pendant stopped working

Hi Andy,
The EStop has two poles. One pole is fed into the MCU of the pendant, which then provides a signal. That is what I have called the "interpreted estop" because it, well, is something that the pendant makes available to LinuxCNC through the USB interface. IMHO, that is not a reliable (i.e. safe) estop.

The second pole is electrically isolated from the pendant electronics (excepting mutual inductance, which we'll get to in a second). The poles are connected to two wires in the shielded cable. The terminus of that cable is an RJ45 connector that splits apart the internal wires into a USB-B connector and two screw terminals. This is all done in something called an EEBox, but that is just a marketing name; in the end, it is just a splitter.

My design for the estop chain in included in an earlier post(forum.linuxcnc.org/10-advanced-configura...g?start=10#gallery-2), but it is fundamentally like what you are describing for your lathe. My VFD/amp power is controlled by a contactor relay with a 120VAC coil. I didn't want that running through my pendant, so instead, the power for that contactor coil is switched by the poles of a small 24VDC coil voltage relay that I call the COR (Control On Relay).
The pendant emergency switch, along with an estop switch on the front panel, that determine whether or not the COR coil is energized, with one minor nuance. There is a feedback in the circuit. While one pole of COR is for the main contactor coil (previous paragraph), the other pole is used within the energizing circuit of the COR relay itself; in essence, it is it's own estop. I did this to allow the use of a momentary estop switch (as well as the fact that I didn't want to reenergize the COR relay without an intentional push of a START button. That START button is a NO switch that is in parallel with the COR pole. You obviously know electronics so I suspect you understand exactly what I am trying to do, but words do a poor job of describing feedback, so if I'm confusing, I'd encourage you to look at the schematic attachment.

OK, that is all how the machine is wired. To eliminate *any* wiring problems, once I decided the issue was likely mutual inductance from the relay coil, I replicated the circuit with a spare relay receptacle that I had. I took an isolated 24VDC 3A supply and basically wired the pendant so that the estop switch acted as a switch to energize this relay... nothing else. Push/release the estop on the pendant and the relay turns on and off. I then added the software hooks for the pendant into the standard LinuxCNC simulator and gave it a whirl. Sure enough, the energizing of the relay can brick the pendant.

Which gets us to your EM pulse.... I concur. This has to be a mutual inductance thing. The cable is shielded, but the estop wires are not separately shielded from the bundle of four USB wires. . The relay is a tiny little OMRON unit. I'm sure you've seen them before because they are pretty common; I can send you the model though if you are interested (don't have it just second).

I can tell you this... I *thought* I had the version of the relay that had a diode incorporated into the unit. I was wrong. I've purchased a new relay to rectify that problem, but in the interim, I added a diode to the terminals of my test harness. It definitely made the problem better, but did not eliminate it altogether. And, there is another very strange thing too... If I remove the relay entirely, there is no current flow when the switch is engaged/disengaged; as such, there is no inductor and, by extension, no EM pulse; it is an open circuit. And, yet, at times, the pendant will still brick. It doesn't happen all the time, but it definitely does do so at times.

All of this is making me think that maybe two tings are going on at the same time. First, the diode definitely helped, so EM seems plausible. Second, it is possible that the second estop pole (the interpreted one) is causing the pendant to brick.... Two transient problems would certainly make it understandable why this has been a pain to diagnose. But, the reverse side is that two transient problems would be fairly rare; the Gods must not like me.

Sorry for the long post.... I'm trying to be thorough but now feel pedantic

I have considered an SSR, but there are three things that have caused me to shy away from them. First, I have that nice DIN mounted relay all wired in and setup. I'm not too sure where I'd put the SSR (yeah, pretty pathetic reason!) . Second, I really want the feedback and I don't think I've ever seen a two pole SSR. Finally, I've had mixed results from SSR relays. For an EStop, I feel more confident using with something that has a real coil and real contacts; it sounds like you echo that thought.

I do agree that an SSR might fix the problem though... I can also try a 5VDC relay to see if that has a different noise characteristic; I doubt it will affect much though. Oh, and yes, I have tried different 24VDC relays and they all cause the same problem.

Yesterday I sent the pendant back to VistaCNC, along with the test power supply, harness, test relay, etc.... Essentially all of what I described in my last post. I'd like to se if they find a problem. Maybe they'll find a way to eliminate the EM noise on the line.

Oh, another option I have is to go to a different cable that has shielded estop leads. It would not be as nice as the coiled wire I have now, though, so that is a downside.... But, I've just started my search for cables; I'm hoping Vista can provide the best solution (one where I don't have to rewire anything other than replace the previously mentioned relays with diode versions... and, yes, I could add the diode to the terminal, but the relays are so cheap that it is easier to just buy new ones).

ON EDIT: I think we've both come to the conclusion about EM. The next curiosity is how that might cause a problem.... USB has four wires: positive, negative, data lines. Because the data lines are differential, they should be more immune to noise or EM. So, my guess is that the power line is where the problem lies. I don't see a restart sequence on the pendant display though...

Did you ever get your problem completely resolved?

Frankly, not really. I ended up not going the SSR route but, instead, bought one of those relay isolator boards on fleaBay. They are cheap and packaged pretty nicely. It uses an opto-isolated input to drive a relay. Being opto-isolated, the power draw through the Vista switch was minuscule, so noise introduced by that transition should have been irrelevant.

By and large, this worked.... but only 98% of the time. Every so often the darned thing would still reset and that was incredibly frustrating. I sent the whole thing back to VistaCNC along with a test rig (cheap power supply and a relay whose coil was switched by an EStop pole (so, the original configuration). The setup failed at Vista too. Their comment was, essentially, that the power supply was not a good supply (as I recall) and that I could change to a different cable. I was kind of hoping they'd swap to a working unit (if it behaved differently), but they didn't. And, yes, the supply was a cheapo thing I bought... just for this test. But I also had the same failure on a high quality bench supply. The Vista guy indicated that it worked on the bench for him, but I don't know how many times he tried. Remember.. better means a lot of switching and then some failures... But those are *still* failures and there shouldn't be any.

So, the long and the short of it is, I had a pendant that worked 98% of the time and 2% of the time caused me to pull my hair out. Worse, when something is transient, it brings into question what else might go wrong. I *love* the idea of the Vista pendant. It is elegant, flexible, and very useful. But I'm not sure I would buy another one of this model, or perhaps any USB type. Plain switches and an MPG are pretty "dumb" but also bullet-proof.

To a greater extent, however, I have moved on. A Dyna DM4400M became available that was completely working. It had better iron than my Hurco. I tried to sell the Hurco with it priced in a way that was essentially a discount on my LinuxCNC parts. I wasn't successful. So, I finally ripped out the control (ouch! that hurt... it was so nice, and almost done) and is now just a collection of parts. I considered taking the electrical box off, but I didn't have the room to store it, and the guy who ended up with the Hurco carcass wanted the box. So (sniff, sniff) I ripped out all of that labor and experimenting.... But! I did learn a lot. And, I have parts for future conversions (maybe even the Dyna if I tire of the Mitsubishi control).

My Vista pendant, along with my Mesa cards, etc, etc, have all been boxed up and shoved on a shelf. I'm not sure what I feel about the pendant. I'd like to use it again because I really liked it, but the fact that I got it to fail on the bench, completely away from any of the stuff I had done, isn't promising.... Switching to a shielded cable *might* make it better, but I don't know for sure.

Why? Are you having the same issue?

I was just reading this old thread and wondered what every happened with your situation. I recently suffered a severe fit of problems. I suspected there might be an issue with the cable as the onset was rapid and severe and the humidity wasn't particularly low which would have had me looking at static discharge. I loosened the gland nut with the strain release on the pendant and pushed the cable in about an inch and lightly snugged the nut. The problem went away for a few days but would then return intermittently. I emailed Lee and he had me send it back I believe they replaced the cable. Very fast service by the way. In any case no problem so far. The activity preceding the failure was different for both of use but they both lead to the same result. In my case I did see three types of failure results. One where the pendant would go back to LINUXCNC on the screen, one where the display and buttons would just freeze and become unresponsive and one where the screen would be garbled. That was the rarest of the three.
One other annoyance was that occasionally Linux would fail to boot but if I unplugged the pendant from the PC and plug it right back in the PC would boot and all would be well. I never really worked to investigate that. It seemed like the OS was waiting for something from the USB and when I unplugged it the boot would continue. I have since updated to Linux Mint 19.3 and LinuxCNC 2.8.0~pre1. VistaCNC has a new vcP4s.hal file for 2.8/2.9. So far so good.

Just updating to this topic, as it has maybe some similarities with my problem.
I had pendant working for a minute and then it crashed. Only Linuxcnc txt on display.
Problem was very unrepeatable, sometimes it did not work at all, sometimes 15min.
I tried it with usb repeater, usb hub, selfmade repeater with -5V input, usb hub with 5v, moving pc far away from control, different usb ports, Bios, updated the linux, ferrites, checking shielding etc etc. very frustrating.
My pendant is p4-Se, with shortest cables.
Last thing I wanted to test was to bypass E-box, that was last becouse proprietary ethernet cable that pendant has, ethernet-usb adapters dont work.
So I chanced whole ethernet cable to USB cable app 2m long. After that there has been NO errors, works like a dream, and liking how the pendant works.
Maybe at some point I will get back figuring hard E-stop, but I think not in a year.

Did you have a arc suppressor across the relay contacts? When switching an inductive load
(like the contactor coil) with a switch or relay, you will generate huge amounts of EMI without
arc suppression (via a snubber for AC coil contactors)

There is no Ethernet involved with the P4-SE. The RJ45 connector on the pendent is used so that it can bring out the USB interface and the E-Stop contacts. The E-Stop contacts are picked off and connected to the screw terminals on the E-box and the USB is passed through to the USB Type B socket. So the E-box screw terminals are just another switch to be placed in the E-Stop chain. That is why the Ethernet-USB adapter didn't work.

Glad you got it working. I really like mine. I'm using the coiled cord. I wish it was wireless but then the E-Stop on the pendent wouldn't be as trust worthy.

At somepoint my pendant would not be regocniced even when the controller box was not powered. So im thinking it wasnt EMI
Maybe the cabels were bad or I have some problems with motherboard. Next i would have bought USB card for a pci slot to test.

My words for Ethernet might have been wrong, should have just used ”Ethernet”

For my machine this 2m cable is totally fine, luckily.